Port ProteoWizard core to .NET 10 - #4619
Open
chambm wants to merge 611 commits into
Open
Conversation
The master merge brought two BiblioSpec C++ changes (the only C++ delta in the merge);
mirror them in the managed port that net8 Skyline actually builds and runs.
BlibMaker.cs (cpp BlibMaker.cpp) - empty-result-build file-lock hardening (a DIA-NN run
that yields 0 precursors aborts the redundant library):
- Init() overwrite: File.Delete is now non-throwing (try/catch), and the existing
File.Exists guard reports the failure message - parity with cpp's error_code bfs::remove.
- AbortCurrentLibrary(): File.Delete is non-throwing and warns instead of throwing out of
abort (parity with cpp's error_code bfs::remove + Verbosity::warn). _db.Dispose() with the
connection's Pooling=false is the managed analog of cpp's sqlite3_close_v2 (release the
file lock before unlinking).
DiaNNSpecLibReader.cs (cpp DiaNNSpecLibReader.cpp) - tolerate DIA-NN 1.9.1's missing Flags
column: read Flags via the nullable FindDataField (optional) instead of MustFindParquetField
(required), defaulting the flags to 0 when absent (the bits aren't consumed downstream). Every
other column stays required via MustFindParquetField, matching cpp's scoped
assert_present_except({"Flags"}).
Verified: BlibBuild builds 0 errors.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…da counts TestDiaTtofDiaUmpireTutorial's audit-log baselines carried the old protein-association counts (Peptides mapped "101", Peptide targets "102") in every language. net8's more conservative OOP-MSAmanda search maps 97 peptides / 98 targets, so the audit-log comparison failed at VerifyAuditLogCorrect - a culture-independent count difference, not a formatting-tolerance case (the .json target counts already pass on net8). Re-recorded the baseline in all five languages (en/fr/ja/tr/zh) via recordauditlogs=on and re-verified that fr passes a normal (non-record) comparison. Only the two counts changed (101->97, 102->98); the localized text is unchanged. Legitimate re-baseline, not a product bug. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Both DiannSearchTest variants failed in the managed BlibBuild's DIA-NN reader (the C# port net8 Skyline builds and runs), from two faithful-port divergences from DiaNNSpecLibReader.cpp introduced with the parquet-report support: 1. TestDiannSearch1_9_1 -- "ERROR: not a parquet file, head: 46696c65": DIA-NN 1.9.1 honours the literal `--out diann-output.parquet` filename Skyline passes (DiannHelpers.cs:717) but writes a TSV report into it. The C# reader dispatched parquet-vs-TSV on the `.parquet` EXTENSION, feeding the TSV to Parquet.Net. cpp routes on CONTENT via ParquetReader::is_parquet (arrow OpenFile().ok(), cpp:731/667). Added an IsParquet() content sniff and routed ReadDiannReport / HasRequiredHeaders / FindLibParquet through it. 2. TestDiannSearch -- "could not find precursorId 'VTHAVVTVPAYFNDAQR3' in speclib": DIA-NN reports precursors absent from the spectral library. cpp SKIPS them at the psmByPrecursorId lookup (cpp:1376-1380, PR #4189) and THROWS only at the entryByModPeptideAndCharge lookup (cpp:1418-1420). The port had these swapped; restored cpp order. Verified on net8: TestDiannSearch + TestDiannSearch1_9_1 both pass (0 failures). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…uites TestPerf and TestTutorial are excluded from the per-commit build.bat (large vendor datasets, long runtimes); tcbuild.bat's scope note calls for a separate build configuration invoking those csprojs directly. tc-perftests.bat is that entry point: it builds Skyline + TestPerf + TestTutorial + TestRunner (plus a best-effort native Hardklor for the feature-detection perf tests), stages them into bin\staging-net8\<Config>, and runs TestPerf.dll and TestTutorial.dll with perftests=on as two TeamCity-decorated suites -- each filtered to its own DLL so the TestFunctional/TestUtil DLLs co-staged as dependencies don't run. Both suites always run so one red suite doesn't hide the other. Mirrors build.bat's flags (--i-agree-to-the-vendor-licenses / --require-vendor-support / --automated / --parallel), the SKYLINE_TEST_ARGS escape hatch, and its restore/build/stage/run helper structure. Staging is done one project per call (Stage-Net8Tests.ps1 via -File binds a comma list as one string), bundling the portable .NET runtime on the first call only. Validated: arg-parse and vendor-guard exit paths; a full end-to-end build -> stage -> run via the escape hatch (TestSciexPrmCeOptimization, 0 failures). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
On the net8 port branch, perf + tutorial tests run via pwiz_tools/Skyline/tc-perftests.bat (dotnet build + Stage-Net8Tests.ps1 + TestRunner perftests=on), not the old cpp/MSVC ProteoWizard_SkylinePrPerfAndTutorialTestsWindowsX8664. Redirect the nightly smart-trigger's master Skyline target to ProteoWizard_SkylineWindowsNetPerfTutorialTests (the net8 config that runs tc-perftests.bat); the old config id is kept commented for restore. Release branch unchanged. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
TestAlphaPeptDeepBuildLibrary compares an AlphaPeptDeep-predicted library against checked-in baselines. peptdeep was pip-installed unpinned (Version=null in AlphapeptdeepLibraryBuilder), so every run pulled the latest peptdeep + pretrained model; newer predictions drift at the 7th significant figure -- enough to reorder near-tied fragments and break the exact baseline comparison. (The test had been labeled "blocked on the Python env", but the net8 embeddable-Python + SkylineProcessRunner harness actually runs the whole install/predict/build pipeline fine in ~3 min; the only failure was this prediction drift, which would fail on net472 identically.) - Pin peptdeep==1.5.0 so the predicted-library baselines are reproducible. NOTE: this freezes the peptdeep version the shipped AlphaPeptDeep feature installs; bump it (and re-record) intentionally. - Re-record with_iRT/without_iRT predict_transformed.speclib.tsv in the test zip from peptdeep 1.5.0 output. Verified on net8: TestAlphaPeptDeepBuildLibrary passes (0 failures). A pinned re-run reproduced the recorded with_iRT baseline exactly (deterministic), confirming the drift was version-driven, not run nondeterminism. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…ine compare TestAlphaPeptDeepBuildLibrary's TestResultingLibByValues sorted the predicted library by (peptide, fragType, fragCharge) but left fragments in AlphaPeptDeep's predicted-intensity order within each group. Predicted intensities differ at the ~7th significant figure across machines (CPU float nondeterminism), which flips near-tied fragments, so the exact positional field compare failed on the CI agent (ProductMz 569 vs 640) even though every value was within the compare's tolerance. Add FragmentSeriesNumber + FragmentLossType to the sort keys so row order is determined by fragment identity, not predicted intensity. The comparison then aligns the same fragment in both files and the tolerance (1.0) absorbs the tiny per-value drift. Validated: the pre-commit baseline (older peptdeep, b9-first) and the peptdeep 1.5.0 baseline (b8-first) -- which differ in fragment order and values like two machines -- compare equal under the new sort (414 rows each, no field diff > tolerance). Local run green. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
… .d handle ConsoleTestDiaPasef / ConsoleTestDiaPasefFullDataset failed on the CI agent at cleanup: Directory.Delete of the imported .d failed with "analysis.tdf is being used by another process". The managed Bruker readers hold raw native handles (tims_open / tsf_open / baf2sql_array_open_storage) and SQLite connections but had NO finalizers, so a reader whose Dispose() was skipped or deferred kept analysis.tdf open until process exit -- and the test cleanup's GC.Collect()+GC.WaitForPendingFinalizers() pass (TestFilesDir.CheckForFileLocks) could not reclaim it. The tests pass locally (this box releases the handle in time); the lock only persists on the CI agent. Give TimsBinaryData, TdfMetadata, TsfBinaryData, TsfMetadata and Baf2SqlData finalizers (standard Dispose(bool) + GC.SuppressFinalize) that release the native handle / SQLite connection, so the cleanup's existing GC pass reclaims a missed/deferred Dispose(). Verified with a throwaway harness: an undisposed TimsBinaryData/TdfMetadata left analysis.tdf locked after GC.WaitForPendingFinalizers() BEFORE this change and released AFTER it. Local ConsoleTestDiaPasef still passes. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The new undisposed-open harness guard (next commit) revealed the same missing-finalizer bug fixed for Bruker in these three readers: each owns a native handle or SDK object that keeps the raw file locked and releases it only on Dispose, with no finalizer -- so a caller who forgets/defers Dispose leaves the file locked past the test cleanup's GC.WaitForPendingFinalizers() pass (the leak that broke Skyline's Bruker .d cleanup on CI). - ThermoRawFile: finalizer disposes the IRawDataPlus / IRawFileThreadManager SDK objects that hold the .raw handle. The finalizer must NOT call GC.WaitForPendingFinalizers() (deadlock on the finalizer thread); only the Dispose() path does. - MobilionData: finalizer frees the native mbi_file handle. Hdf5Lock is a thread-affine Monitor owned by the Open() thread, so Monitor.Exit is skipped on the finalizer path (the native handle -- what actually locks the .mbi -- is freed regardless). - WiffFile (legacy .wiff): finalizer calls AnalystWiffDataProvider.Close() (+ sample / experiment disposes). Non-SIM/SRM samples release on Close; SIM/SRM stay soft-failed. Verified: Bruker/Thermo/Sciex/Mobilion vendor tests pass the undisposed-open guard; Waters/Agilent/Shimadzu/UNIFI/UIMF were already clean. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…r harness The disposed-path AssertFilesUnlocked probe verifies a DISPOSED reader releases the raw file, but can't catch a reader whose backing data has no finalizer -- a caller who skips/defers Dispose then leaks the handle (the class of bug that broke Skyline's Bruker .d cleanup on CI while the vendor harness stayed green). Add ProbeFinalizers: for each fixture, open it through the reader, drop the reference WITHOUT Dispose, force GC + finalizers, and assert the raw file can be renamed. Runs from FixtureRunContext.Check() -- AFTER all config runs -- because a genuinely missing finalizer leaks a handle that would otherwise cascade into the disposed-path probes of the remaining configs. Honours the same IsKnownLeakySdkPath soft-fail (wiff2 / wiff+srm-sim are ALC-limited). This guard found the missing finalizers now fixed in Thermo, Sciex, and Mobilion (previous commit), in addition to Bruker. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Temporary, to be reverted. Adds a diagnostic to TestDiannSearchTutorial that dumps the chromatograms-page results-file match state (doc-library sources, FOUND Name=>Path, MISSING SpectrumSourceFiles keys) and fails with it in the message when ResultsFilesMissing. Also scopes tc-perftests.bat to run only TestDiannSearchTutorial, to capture the CI-only missing-key list cheaply. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
Extends the temporary diagnostic to enumerate the persistent CacheDir and DocDir so the CI failure shows any stale/foreign .quant (e.g. _03 replicates) and stale diann-output* artifacts feeding the blib's 6 unfindable .raw sources. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
…to searched files Candidate fix. PrepareDocument kept diann-output* (report/output-lib/converted speclib) across runs in the stable DocDir. On a CI agent whose persistent DiannPerfDoc-stable was populated by an earlier run over the full 6-file ProteoBench dataset (runs named after the original .raw acquisitions), those stale outputs leaked 6 .raw source names into the freshly built document library, so the chromatograms page could not find any of the 4 searched .half.mzML files (ResultsFilesMissing) -> "Some results files are still missing" -> TestDiannSearchTutorial failed on CI (passed locally, where the DocDir only ever held 4-mzML outputs). Now also wipe diann-output* each run, keeping the predicted library + .quant cache. Diagnostic + tc-perftests scope still present to confirm on CI; both to be reverted next. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
…ributed to searched files" This reverts commit 93e296d.
The DIA-NN intermediates (output report + output library + its Skyline-converted speclib + the .sky/.blib) lived in a stable per-user dir (DiannPerfDoc-stable). DIA-NN's output library is reused by the --use-quant flow and never rebuilt, so once that stable dir was populated by a run over a different file set (the full 6-file ProteoBench dataset, whose runs are named after the original .raw acquisitions) every later run reused it and built a document library attributed to 6 unfindable .raw sources -> "Some results files are still missing" -> TestDiannSearchTutorial failed on that CI agent (passed elsewhere). Move DocDir to the normal per-test results location (TestContext.GetTestResultsPath), fresh each run, so no stale library can persist. The fast-run cache (per-file .quant + predicted spectral library) stays in the persistent CacheDir. Diagnostic + tc-perftests scope still present to confirm on CI; both reverted next. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
The DocDir move (per-test ephemeral results dir) is confirmed green on CI: the scoped run (build #66, TestDiannSearchTutorial-en) passed in 1025s with the document library now attributed to the 4 searched files. Restore the diagnostic block in ImportToSkylineDocument and the SKYLINE_TEST_ARGS scoping in tc-perftests.bat to their originals. Net change of this session vs the pre-diagnostic head is just the DocDir move. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
…-15 demux re-baseline) Commit 993b3d9 (2026-07-15) re-baselined FinalTargetCounts for the net8 C#-ported NNLS DIA demultiplexer (369->362) but left two other net472-recorded baselines stale, which only surfaced once the earlier assert was fixed: - MassErrorStats: the {-0.2,2.5} means were recorded on native msconvert (2025-08). net8's demux shifts the wide-window panes; #if-split to net8 {-0.3},{-0.3},{-0.2} (stdDevs stay 2.5). Values are deterministic (bit-stable across runs, unchanged with hardware intrinsics disabled) -- not machine-dependent. - Tutorial audit logs (en/fr/ja/tr/zh): still carried the net472 counts (Proteins mapped 369, Peptides 678, ...); re-recorded for net8 (362/648/...). Verified green: TestEncyclopeDiaSearchTutorialDraft (en) 0 failures. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
Env-gated (SKYLINE_MSAMANDA_DUMP_DIR) copy of each file's Percolator input (<base>_pin.tsv, gzipped) and mzIdentML search output (.mzid.gz) so a CI-agent run and a local run can be diffed file-for-file. Localizes whether the +/-2 LibraryPeptideCount drift originates in the MSAmanda search (pin features differ across CPUs) or the embedded Percolator FDR (pin identical, q-values flip). tc-perftests.bat scopes the run to TestDiaTtofDiaUmpireTutorial and publishes the dump as an artifact even when the count assert fails. Revert after capture. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
The first capture attempt published nothing: the dump dir under TestResults did not exist at publish time on the CI agent (env var may not reach the test process, and/or cleanup=all wipes TestResults). Make it robust: - dump OUTSIDE TestResults (SCRIPT_DIR\msamanda_dump), cleanup-proof - after the run, capture-msamanda-dump.ps1 finds the *-diaumpire_pin.tsv the test always leaves in the downloads dir (independent of the wrapper), copies recent ones to the dump dir, and logs exactly what/where (CAPTURE: prefix) - publish unconditionally Revert after capture. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
The wrapper dump worked (4 files, 172MB) once moved outside TestResults, but the artifact was still empty: `echo ##teamcity[publishArtifacts '... => file.zip']` has a '>' in '=>' which cmd treats as redirection, so the service message was written to a file instead of emitted. Publish the dump directory directly (no '=>'). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
Proof-of-concept: switch TestDiaTtofDiaUmpireTutorial from MSAmanda to Comet to eliminate the per-machine +/-2 LibraryPeptideCount drift at the root. Empirically, MSAmanda's scores are machine-identical but it writes PSMs in thread-completion order and resolves the isobaric I/L tiebreak by protein-encounter order -- both vary with core count, shifting a few I/L string variants across the q<=0.01 cutoff. Comet's pin is byte-identical across thread counts (content + order), so its result is machine-stable. - select SearchEngine.Comet, FragmentIons "b,y", high-res MS2 analyzer, no fragment tol - re-record TTOF baselines for Comet (LibraryPeptideCount 12221 -> 13822, iRT, mass errors) - broaden CleanUpPersistentDir to remove Comet/crux outputs (-percolator.pepXML, comet.*, make-pin.pin, percolator.*) so the persistent-dir-modified guard passes - audit-log comparison temporarily disabled (workflow steps change vs recorded MSAmanda logs) Passes locally (0 failures). Evaluating cross-machine determinism on CI next. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
The Comet PoC CI build failed to compile with CS0579 'Duplicate AssemblyVersionAttribute' in Skyline's generated obj/Net8Version.g.cs -- a stale obj left by a concurrent build sharing the agent's C:\pwiz checkout (unrelated to the test change; builds clean locally). Remove Skyline's obj before restore so the version file regenerates cleanly. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
CometSearchEngine set spectrum_batch_size=0 (load and search all spectra at once), apparently copied from Comet's stock default. For large inputs -- notably DIA-Umpire pseudo-spectra (~150k+ spectra per file) -- all-at-once is ~2.5-4x slower than batching due to memory/cache pressure. Use crux's default batch of 20000. Purely a performance knob; search results are identical (verified: same LibraryPeptideCount). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
Removes the diagnostics used to localize the DiaUmpire per-machine drift: - MSAmandaSearchWrapper.DumpSearchArtifacts (env-gated pin/mzid dump) - capture-msamanda-dump.ps1 - tc-perftests.bat hooks: the hardcoded `set SKYLINE_TEST_ARGS=test=...` scoping, the SKYLINE_MSAMANDA_DUMP_DIR/publishArtifacts capture, and the obj clean. Importantly this restores tc-perftests.bat to running the FULL TestPerf + TestTutorial suites: the hardcoded scoping had (since d3dc8ad) limited every perf CI run to a single test. The SKYLINE_TEST_ARGS escape hatch remains available as an external env var for scoped runs. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
MSAmanda's parallel-order isoleucine/leucine tiebreak produced a machine-dependent +/-2 peptide drift in the DiaUmpire tutorial tests. Switch the search to Comet (equal_I_and_L=1), which is byte-for-byte deterministic across machines. - Wire Comet with num_output_lines=1 (small Percolator pin, fast FDR). - Manual-review click: a Comet peptide can carry multiple precursor charges, which the chromatogram graph draws as per-precursor TOTAL curves that have no full-scan info, so the full-scan tracking dot never appears. Select a single precursor (+ ShowAllTransitions) before clicking so individual transition curves display. Retune the QE click point to Comet's apex (must land within CheckFullScanSelection's 10px tolerance of a real transition vertex). - Re-record all four Comet baselines and the QE/TTOF en audit logs. QE-Extra LibraryPeptideCount 7574->8742; both CI-run variants (TestDiaQeDiaUmpireTutorialExtra, TestDiaTtofDiaUmpireTutorial) validate green in CI config. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
The net8 pwiz_tools/Skyline/tc-perftests.bat never applied the long-running-test skiplist (the legacy scripts/misc/tc-perftests.bat did it via findstr; the net8 rewrite dropped it), so the perf/tutorial config ran ~100 min of tests the PR perf agent skips and hit the execution timeout. Pass the skiplist to TestRunner via its native skip=@file. - Move tc-perftests-skiplist.txt next to the net8 runner in pwiz_tools/Skyline/ (mirrors the earlier tc-perftests.bat move). - Delete the superseded legacy scripts/misc/tc-perftests.bat (dead on net8: it invoked the old bin\x64\Release TestRunner path). - Sync the list with master (add the disk-heavy TestDiaToSrmTutorial). Verified with TestRunner listonly: skiplisted tests are excluded, the DiaUmpire tutorial test still runs. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
The en-only Comet re-record left the fr/ja/tr/zh DiaUmpire tutorial audit logs on the old MSAmanda content (128 lines vs en's 132), which the CodeInspection cross-language line-count check flags. Re-recorded the other four languages for both DiaUmpire tutorial tests (recordauditlogs) so every language matches en at 132 lines. New Comet-workflow strings not yet translated appear in English until localized (standard). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
…onfig net472's perf/tutorial config runs the tutorials in both English (Pass 2) and French (Pass 0: French number format, mzML, no vendor readers, no internet); the net8 tc-perftests.bat ran English only, so ~26 French tutorial tests weren't covered. Add pass0=on to the TestTutorial invocation to restore the French pass. TestRunner auto-skips perf tests in Pass 0, so this applies only to the standard tutorials; the perf suite still runs English-only, matching net472. Verified locally: TestMethodEditTutorial runs green in French Pass 0. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
Three perf tests failed on net8 (pass on net472) because the managed pwiz-sharp readers dropped code paths the native readers have: - Bruker TDF combined-IMS spectra emitted inconsistent native ids: the index entry was "merged=N" but the emitted spectrum was "merged=N frame=F scanStart=S scanEnd=E". C++ uses the 4-field form for both, so the abbreviated scan id (N.F.S.E) round-trips through FindAbbreviated (needed by ion-mobility library population). Build the 4-field id in MakeCombinedEntry and drop the redundant FillSpectrum override. Makes CommandlineCreateImsDbPerfTest build the imsdb (its one remaining failure is an unrelated net8 Path.GetFullPath leniency in a negative test case, not the reader). - Agilent AgilentRawData threw when MassSpecDataReader.OpenDataFile returned false; C++ MassHunterData.cpp ignores it -- the false only flags a possibly-incomplete acquisition (e.g. a "[stitch]" multi-CE .d), the reader is still usable. Fixes AgilentIMSImportFringeValuesTest. - Waters SONAR: the ReportSonarBins flag was never threaded into the reader, so the combined SONAR path emitted quadrupole m/z instead of the SONAR bin number in the lower-bound array and Skyline's bin filter matched nothing. Thread the flag through and emit the drift-scan bin, mirroring SpectrumList_Waters.cpp. Fixes WatersSonarPerfTest. Agilent and Waters SONAR verified green by running their perf tests. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
…gent The perf/tutorial config (ProteoWizard_SkylineWindowsNetPerfTutorialTests) shares its checkout on the MacCoss agent, so a concurrent build can leave a stale generated Net8Version.g.cs in Skyline's obj. Compiled alongside the copy stamped by Skyline.csproj's StampNet8InformationalVersion target it produces CS0579 'Duplicate AssemblyVersionAttribute' and fails the build before any test runs (builds clean locally). rmdir Skyline's obj at the start so the version file regenerates from a clean incremental state -- cheap next to the hours-long test run. (A temporary form of this was in 53deca6 and got reverted with unrelated diagnostics.) Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R1mZ7FBbk2WjDNsio3v1io
…0818_commonutil_winforms_split
* 2cb66ee moved global.json under pwiz-sharp because the root pin was then 8.0.100, which capped the whole repo at C# 12 and broke Osprey's publish * b882847 retargeted that pin to 10.0.100, so a root pin now grants C# 14 instead of capping it and the reason for the split is gone * Left where it was, the root resolved to whatever SDK the agent happened to have - C# 12 on an 8.x-only agent - so this also makes the band deterministic * Contents are unchanged; the original move was a pure rename (R100) Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015DtKunDbAtAsmUyx9uGnJg
* dotnet --list-sdks prints "8.0.423 [/usr/share/dotnet/sdk]", and an unescaped ] ends a service message, so build #156 cut the warning at the first bracket and logged a parse error instead of the SDK list it exists to report * Added tc_escape - | doubled first, then ' [ ] and newline - and four cases pinning it, since a diagnostic only fires when something is already wrong Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015DtKunDbAtAsmUyx9uGnJg
…onutil_winforms_split skyline: Split CommonUtil's WinForms half so ProteowizardWrapper can be plain net8.0
* Baf2Sql.cpp's copy of translateInstrumentFamily was missing case 9; the TDF and TSF copies have it, and the function is per-translation-unit * A timsTOF can write BAF, so through that path the instrument resolved to Unknown and emitted MS:1000122 with a source-only componentList * Pinned the code table in BrukerInstrumentFamilyTests, which needs no vendor SDK, so the C# side cannot be "corrected" back to cpp's omission Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015DtKunDbAtAsmUyx9uGnJg
* GetSpectrumByRow passed null peak filters, which tells MHDAC not to centroid rather than not to filter, so PeakElseProfile returned the stored profile * Only bites a file storing profile with no peaks; anywhere peaks are stored the SDK returns those and the missing filter changes nothing * Added Neg_MS_002_1scan.d, a first-scan subset of the corpus file, as the fixture for that case, with default and vendor-centroid references * Reverting the fix makes the fixture fail on the centroid reference Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015DtKunDbAtAsmUyx9uGnJg
* Agilent/Sciex/Shimadzu/UIMF/Mobilion were excluded from build.sh entirely, so nothing checked that those readers still identify their formats off-Windows * VendorReaderTestHarness already converts VendorSupportNotEnabledException into an Identify() assertion, the runtime equivalent of cpp's testAcceptOnly, so the suites needed adding to the list and nothing else * Ungated by the licence flag, since identify-only needs no vendor SDK * Two tests read through the reader directly rather than the harness, so they now report Inconclusive instead of letting the exception escape Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015DtKunDbAtAsmUyx9uGnJg
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
Core Linux .NET went red on 4387ba6 before running a single test: Sciex.Tests.csproj(68,5): error MSB3030: Could not copy the file ".../pwiz-sharp/vendor-assemblies/Sciex/SQLite.Interop.dll" because it was not found. StageWiff2NativeAssembliesForTests was gated on $(IAgreeToVendorLicenses), which the Linux CI passes, rather than on $(NativeVendorsAvailable), which is that AND Windows. The files it stages come from Sciex.csproj's archive extraction, and that only runs on Windows - vendor-assemblies/ is gitignored, so on a Linux agent it is simply absent. The target therefore ran where its inputs could not exist and failed the build, taking all 389 tests with it. MsConvert.csproj's StageWiff2NativeAssemblies already carries this exact reasoning in a comment, and every other consumer of $(PwizVendorAssembliesPath) uses $(NativeVendorsAvailable). This target was the lone outlier. Verified on Windows that nothing changes: deleted the staged wiff2 directory, rebuilt, and it was recreated with SQLite.Interop.dll, System.Data.SQLite.dll, Unity.Abstractions.dll and the CRT, 0 errors. Verified on Linux in a dotnet sdk:10.0 container that the build now succeeds. The CI failure is the reproduction; it cannot be reproduced in that container, because the mounted host tree carries a vendor-assemblies/ populated by a Windows build, which is precisely the file a fresh Linux agent does not have. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0182oRXbeKQGEpF1kTkdQAsr
…rcase files
Core Linux .NET failed Reader_Agilent_Neg_MS_002_1scan_ProfileOnlyVendorCentroid:
InvalidDataException: Not an Agilent .d directory: .../Neg_MS_002_1scan.d
IsAgilentDirectory probed for AcqData/MSScan.bin and AcqData/MSPeak.bin with
File.Exists, which follows the platform and so is case-sensitive on Linux. That
fixture stores msscan.bin in lowercase - it is the only one of the twelve
Agilent fixtures that does, the other eleven ship MSScan.bin - so the directory
was rejected there while working on Windows.
Not just a test-data quirk: the on-disk case comes from the acquisition
software, so any real .d written in lowercase was unreadable on Linux. Identify()
uses the same helper, so it returned CVID_Unknown for those directories too, and
the harness's identify-only fallback could not rescue them either.
Switched to EnumerationOptions.MatchCasing.CaseInsensitive, which states the
casing rather than inheriting MatchCasing.PlatformDefault. BrukerData already
handles this for its own filenames by probing two spellings; Agilent was the
one reader still assuming the platform.
Verified on a genuinely case-sensitive filesystem (a container's own fs, not a
bind mount) that File.Exists("MSScan.bin") is False against an on-disk
msscan.bin while the new probe is True, and that MSPeak.bin stays False so the
check does not widen. Agilent.Tests is 17/17 on Windows.
Worth recording for anyone reproducing Linux behaviour here: a Docker Desktop
bind mount of a Windows checkout inherits NTFS case-insensitivity, so the
container finds MSScan.bin and this bug cannot reproduce over that mount. It
needs the container's own filesystem, or a real Linux agent.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0182oRXbeKQGEpF1kTkdQAsr
The disabled TeamCity "Install dotCover (.NET)" steps are deliberate - steps belong in the build scripts - but nothing had taken over the job for Osprey, so its build.ps1 still required a global `dotcover` on PATH and exited 2 without one. That held only on MacCoss Agent 1, which happens to have it; three different ephemeral AWS agents failed identically. Added pwiz-sharp/scripts/Ensure-DotCover.ps1: restores dotCover from a given directory's .config/dotnet-tools.json and proves the tool runs before anything depends on it. Core and Osprey both call it, Core replacing the equivalent inline restore it already had. A local manifest rather than `dotnet tool install -g`, following what Core already did: the version is pinned in source and the build leaves nothing on the agent. Also, the package the old message named does not exist - JetBrains.dotCover.GlobalTools is not on nuget.org at all, the package is JetBrains.dotCover.CommandLineTools - so anyone following that instruction failed. Corrected in build.ps1 and tcbuild.bat. The pinned version stays per-app, because the callers are not interchangeable: Skyline's TestRunner resolves ...\2023.3.3\tools\dotCover.exe by path and passes the old /Filters= syntax, while Osprey passes the kebab-case flags that replaced it. Osprey therefore pins 2026.1.1, the version its arguments were written against. Two things that only showed up by trying it: the package's command name changed (dotnet-dotCover in 2023.3.3, dotCover in 2026.1.1), so the manifests differ there too; and `--version` is not a usable liveness probe, since 2023.3.3 exits 127 on it while 2026.1.1 exits 0 - `help` is 0 on both, so the shared script uses that. Osprey now invokes `dotnet dotcover` rather than a global exe, so it gets the pinned version. Skyline is left alone: it has the manifest, but --coverage in its build.bat is deliberately disabled and warns as much, so wiring a restore there would be dead code until someone re-enables it. Verified: the shared script restores and probes cleanly against all three manifests (Osprey 2026.1.1, Core and Skyline 2023.3.3), and both edited PowerShell files parse clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0182oRXbeKQGEpF1kTkdQAsr
…g directory My previous commit broke Osprey's coverage run: 1c28a26 invoked the restored tool as `dotnet dotcover`, and TeamCity runs this build from the repo root, where Could not execute because the specified command or file was not found. * You intended to execute a .NET program, but dotnet-dotcover does not exist. A local tool is only on the command line when the working directory is at or under its manifest, and the manifest is in pwiz_tools\Osprey. I had verified the command from inside that directory, which is not how the build calls it, so the check passed locally and failed in CI. Moving the working directory would not be an acceptable fix either - it would change CWD for the tests running underneath. Ensure-DotCover.ps1 now returns the launcher path, which works from anywhere, and Osprey uses it as the process filename as it did before this pair of commits. The two pinned generations package the runner differently, which the first attempt also missed: 2023.3.3 ships tools\dotCover.exe, 2026.1.1 ships a managed tools\dotCover.dll with no launcher at all. The script returns whichever exists and the caller runs a .dll through `dotnet`. Verified from the repo root this time, which is the condition that failed: resolution returns dotCover.dll for Osprey (2026.1.1) and dotCover.exe for Core (2023.3.3), both paths exist, and both launch forms exit 0 when run from there. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0182oRXbeKQGEpF1kTkdQAsr
Both Core and Osprey ended their coverage runs with "No available .NET Coverage report generator for type 'dotcover'" and the imported .dcvr skipped, so no Code Coverage tab was produced. TeamCity's agent-side processor needs a dotCover of its own to render the snapshot, and that used to come from the "Install dotCover (.NET)" build step; now that the steps live in the build scripts, nothing provisions it. Ensure-DotCover.ps1 emits ##teamcity[dotNetCoverage dotcover_home=...] pointing at the copy it just restored, which both builds pick up for free. That home has to contain dotCover.exe, which the 2026.1 console runner no longer ships (managed dotCover.dll only, and no HTML report type either). Osprey was the one app pinned to 2026.1.1, so it moves to the 2023.3.3 that Core and Skyline already use and its cover invocation goes back to the /PascalCase syntax that runner speaks; the exclusion list is translated verbatim into -:module= filters, so the same assemblies are measured. Verified locally: 592 tests under coverage on net10.0, 46.2% of 34361 statements across Osprey.* plus pwiz.CommonUtil, no third-party modules. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0182oRXbeKQGEpF1kTkdQAsr
brendanx67
changed the title
3 tasks
…eak (#4629) * Fixed eight net10 nightly failures that were one defect: the benign ObjectDisposedException WinForms raises when signalling a marshaled call whose wait handle it already disposed (dotnet/winforms#14996) * Fixed the pass-1 managed leak from undisposed NHibernate SessionFactories in IonMobilityDb and OptimizationDb, and kept the factory owned on the throw paths * Fixed TestAuditLogTutorial waiting on a document grid that was not open yet * Stopped a superseded loader's self-cancellation ending a results test's wait * Added a loader lifecycle trace, since the thread dump at that failure is always empty of loader threads See TODO-20260818_commonutil_winforms_split.md in pwiz-ai/todos Co-Authored-By: Claude <noreply@anthropic.com>
3 tasks
8 tasks
…izard binaries matching the active configuration (#4634) * Skyline.csproj references pwiz-sharp projects outside Skyline.sln, so MSBuild stripped Configuration/Platform and they fell back to Debug/AnyCPU * Added an AssignOutOfSolutionProjectReferenceConfiguration target in pwiz_tools/Directory.Build.targets that gives unassigned project references the solution Configuration and Platform=AnyCPU * Passed ShouldUnsetParentConfigurationAndPlatform=false as a global property so the fix flows down the whole out-of-solution subtree * Conditioned on CurrentSolutionConfigurationContents, so command-line csproj builds via build.bat are unaffected See TODO-20260612_net8_port.md in pwiz-ai/todos Co-Authored-By: Claude <noreply@anthropic.com>
Master now carries the IdentityModel 7 work (#4637, merged as 79454aa), so where both sides had independently solved the same problem this takes master's shared implementation and keeps the branch's net8-specific knowledge. Auth: both sides replaced the TokenClient constructors IdentityModel 7 removed, the branch with the RequestPasswordTokenAsync/RequestRefreshTokenAsync extensions, master with OAuthPasswordGrantClient. Master's wins because the native C++/CLI UNIFI/waters_connect readers call the same helper, and it drops the IHttpClientFactory pipeline and the two mock handler classes master deleted. The branch's Task.Run existed to keep the async token call off the UI thread; the shared helper's request is synchronous, so nothing regresses. Kept the branch's comments on why IdentityModel 7 populating TokenResponse.Raw makes the invalid_scope and invalid_grant branches reachable, minus the net472-versus-net8 contrast that no longer holds now master is on IdentityModel 7 too. UnifiFunctionalTest: took master's graph handling, which tiles the graphs and asserts an explicit per-replicate curve count, over the branch's SelectedGraphChromName approach; the tiling call had already merged into the surrounding code. The branch's socket-derived ConnectionRefusedMessage and DnsResolutionFailedMessage are outside the conflicts and survive - they are required, since .NET 8 surfaces the raw Winsock text where net472 raised a WebException. Master also guards the invalid-password assertion behind the waters_connect check, because that message is Waters wire text a Unifi server need not match. csproj: kept the branch's SDK-style projects, whose globs cover master's explicit Compile and EmbeddedResource items, and ported the one semantic addition - the IdentityModel reference CommonUtil now needs for OAuthPasswordGrantClient. Master's Microsoft.Bcl.AsyncInterfaces pin is not carried: it exists for net472 app.config binding redirects and CommonMsData is net10.0-windows only here. .gitignore keeps the branch's unanchored entries and adds master's rolled-log pattern. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0182oRXbeKQGEpF1kTkdQAsr
chambm
added this pull request to stack #4647
September 9, 2026 17:25
chambm
marked this pull request as ready for review
September 9, 2026 17:28
There was a problem hiding this comment.
🟡 Changes recommended
There is at least one verified runtime-breaking issue (opening the .NET download URL without UseShellExecute on .NET Core+) plus required style/localization guideline violations in changed code.
Once you've addressed the issues Copilot identified, you can request another Copilot review.
Pull request overview
Ports ProteoWizard’s managed stack forward to .NET 10 and integrates the new pwiz-sharp/ C# core port alongside broad Skyline/Osprey/test updates needed for runtime/SDK behavioral differences (process launching, WinForms dialog behavior, float formatting determinism, test infrastructure).
Changes:
- Add and wire up
pwiz-sharp/(managed ProteoWizard core + tools/tests scaffolding) and supporting build tooling. - Update Skyline, Shared, and Osprey code/tests for .NET 10 runtime differences (shell launching, folder picker parity, formatting determinism, API changes).
- Repository/build hygiene updates (global SDK pin, git attributes/ignores, TeamCity path triggers, tool manifests).
File summaries
| File | Description |
|---|---|
| scripts/test/untar_test_data.bat | Batch formatting/line-ending related adjustment |
| scripts/test/tar_test_data.bat | Batch formatting/line-ending related adjustment |
| scripts/misc/show_boostver.bat | Batch formatting/line-ending related adjustment |
| scripts/misc/nightly_trigger_and_paths_config.py | Nightly trigger switched to net build config |
| pwiz/data/vendor_readers/Waters/generate_waters_mzml.bat | Batch formatting/line-ending related adjustment |
| pwiz/data/vendor_readers/UIMF/generate_uimf_mzml.bat | Batch formatting/line-ending related adjustment |
| pwiz/data/vendor_readers/Thermo/generate_thermo_mzml.bat | Batch formatting/line-ending related adjustment |
| pwiz/data/vendor_readers/Shimadzu/generate_shimadzu_mzml.bat | Batch formatting/line-ending related adjustment |
| pwiz/data/vendor_readers/Bruker/generate_bruker_mzml.bat | Batch formatting/line-ending related adjustment |
| pwiz/data/vendor_readers/Agilent/Reader_Agilent_Test.data/Neg_MS_002_1scan.d/AcqData/msts.xml | Add Agilent fixture file |
| pwiz/data/vendor_readers/Agilent/Reader_Agilent_Test.data/Neg_MS_002_1scan.d/AcqData/interactive.m/info.xml | Add Agilent fixture file |
| pwiz/data/vendor_readers/Agilent/Reader_Agilent_Test.data/Neg_MS_002_1scan.d/AcqData/Devices.xml | Add Agilent fixture file |
| pwiz/data/vendor_readers/Agilent/Reader_Agilent_Test.data/Neg_MS_002_1scan.d/AcqData/Contents.xml | Add Agilent fixture file |
| pwiz/data/vendor_readers/Agilent/Reader_Agilent_Test.data/Neg_MS_002_1scan.d/AcqData/acqmethod.xml | Add Agilent fixture file |
| pwiz/data/vendor_readers/Agilent/generate_agilent_mzml.bat | Batch formatting/line-ending related adjustment |
| pwiz/data/vendor_readers/ABI/generate_abi_mzml.bat | Batch formatting/line-ending related adjustment |
| pwiz-sharp/Tools/SeeMS/src/Misc/SpectrumDataSet.xss | Add SeeMS designer layout artifact |
| pwiz-sharp/Tools/SeeMS/src/Misc/SpectrumDataSet.xsc | Add SeeMS designer settings artifact |
| pwiz-sharp/Tools/SeeMS/src/Misc/SpectrumDataSet.cs | Add SeeMS dataset partial/constants |
| pwiz-sharp/Tools/SeeMS/src/Misc/ChromatogramDataSet.xss | Add SeeMS designer layout artifact |
| pwiz-sharp/Tools/SeeMS/src/Misc/ChromatogramDataSet.xsc | Add SeeMS designer settings artifact |
| pwiz-sharp/Tools/MsConvertGUI/src/Properties/Settings.settings | Add MsConvertGUI user settings scaffold |
| pwiz-sharp/Tools/MsConvertGUI/src/CustomDataSourceDialog/Properties/Settings.settings | Add CustomDataSourceDialog user settings scaffold |
| pwiz-sharp/Tools/Commandline/MsConvert/test/Usings.cs | Add global test usings |
| pwiz-sharp/Tools/Commandline/MsConvert/test/MsConvert.Tests.csproj | Add MsConvert test project |
| pwiz-sharp/Tools/BiblioSpec/test/BiblioSpec.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/Tools/BiblioSpec/src/BlibToMs2/BlibToMs2.csproj | Add BlibToMs2 tool project |
| pwiz-sharp/Tools/BiblioSpec/src/BlibSearch/BlibSearch.csproj | Add BlibSearch tool project |
| pwiz-sharp/Tools/BiblioSpec/src/BlibFilter/BlibFilter.csproj | Add BlibFilter tool project |
| pwiz-sharp/scripts/flatten-tool.sh | Add repo maintenance helper script |
| pwiz-sharp/README.md | Add pwiz-sharp overview docs |
| pwiz-sharp/pwiz/test/Waters.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/Util.Tests/Util.Tests.csproj | Add Util test project |
| pwiz-sharp/pwiz/test/Util.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/Util.Tests/SmokeTest.cs | Add basic assembly smoke test |
| pwiz-sharp/pwiz/test/Util.Tests/FloatingPointTests.cs | Add floating-point behavior tests |
| pwiz-sharp/pwiz/test/UNIFI.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/UIMF.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/TraData.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/TraData.Tests/TraData.Tests.csproj | Add TraData test project |
| pwiz-sharp/pwiz/test/Thermo.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/Shimadzu.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/Shimadzu.Tests/Shimadzu.Tests.csproj | Add Shimadzu test project + BinaryFormatter opt-in |
| pwiz-sharp/pwiz/test/Sciex.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/MsData.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/MsData.NativeAot.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/Mobilion.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/Mobilion.Tests/Mobilion.Tests.csproj | Add Mobilion test project |
| pwiz-sharp/pwiz/test/IdentData.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/IdentData.Tests/IdentData.Tests.csproj | Add IdentData test project |
| pwiz-sharp/pwiz/test/Common.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/Common.Tests/SmokeTest.cs | Add basic assembly smoke test |
| pwiz-sharp/pwiz/test/Common.Tests/Common.Tests.csproj | Add Common test project |
| pwiz-sharp/pwiz/test/Bruker.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/Bruker.PrmScheduling.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/Analysis.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/test/Analysis.Tests/SpectrumProcessing/SpectrumList_ChargeFromIsotopeTest.data/case10.intensity.txt | Add analysis test fixture |
| pwiz-sharp/pwiz/test/Agilent.Tests/Usings.cs | Add global test usings |
| pwiz-sharp/pwiz/src/Vendor/UNIFI/UnifiReaderRegistration.cs | Add vendor reader registration helper |
| pwiz-sharp/pwiz/src/Vendor/Shimadzu/ShimadzuReaderRegistration.cs | Add vendor reader registration helper |
| pwiz-sharp/pwiz/src/Vendor/Bruker/BrukerReaderRegistration.cs | Add vendor reader registration helper |
| pwiz-sharp/pwiz/src/Vendor/Bruker.PrmScheduling/MeasurementMode.cs | Add Bruker PRM scheduling managed wrapper |
| pwiz-sharp/pwiz/src/Vendor/Bruker.PrmScheduling/AdditionalMeasurementParameters.cs | Add Bruker PRM scheduling managed wrapper |
| pwiz-sharp/pwiz/src/Util/Util.csproj | Add core Util project |
| pwiz-sharp/pwiz/src/Util/PhysicalConstants.cs | Add chemistry constants |
| pwiz-sharp/pwiz/src/Util/MassAbundance.cs | Add isotope record struct |
| pwiz-sharp/pwiz/src/Util/AssemblyInfo.cs | Add Util assembly marker/metadata |
| pwiz-sharp/pwiz/src/TraData/TraData.csproj | Add TraData project |
| pwiz-sharp/pwiz/src/TestHarness/TestHarness.csproj | Add vendor test harness project |
| pwiz-sharp/pwiz/src/MsData/Sample.cs | Add MSData sample model |
| pwiz-sharp/pwiz/src/MsData/ProteoWizardVersion.cs | Add MSData version helper |
| pwiz-sharp/pwiz/src/IdentData/SpectrumIdentificationResult.cs | Add IdentData schema model |
| pwiz-sharp/pwiz/src/IdentData/SpectrumIdentificationList.cs | Add IdentData schema model |
| pwiz-sharp/pwiz/src/IdentData/SpectraData.cs | Add IdentData schema model |
| pwiz-sharp/pwiz/src/IdentData/SequenceCollection.cs | Add IdentData schema model |
| pwiz-sharp/pwiz/src/IdentData/Peptide.cs | Add IdentData schema model |
| pwiz-sharp/pwiz/src/IdentData/IdentifiableParamContainer.cs | Add IdentData base type |
| pwiz-sharp/pwiz/src/IdentData/Identifiable.cs | Add IdentData base type |
| pwiz-sharp/pwiz/src/IdentData/IdentData.csproj | Add IdentData project |
| pwiz-sharp/pwiz/src/IdentData/DataCollection.cs | Add IdentData schema model |
| pwiz-sharp/pwiz/src/IdentData/AnalysisData.cs | Add IdentData schema model |
| pwiz-sharp/pwiz/src/Common/IIndex.cs | Add common indexing abstraction |
| pwiz-sharp/pwiz/src/Common/Common.csproj | Add Common project |
| pwiz-sharp/pwiz/src/Common/AssemblyInfo.cs | Add Common assembly marker/metadata |
| pwiz-sharp/pwiz/src/Analysis/PeakPicking/ISmoother.cs | Add Analysis interface |
| pwiz-sharp/pwiz/src/Analysis/PeakPicking/IPeakDetector.cs | Add Analysis interface/types |
| pwiz-sharp/pwiz/src/Analysis/PeakFilters/ISpectrumDataFilter.cs | Add Analysis interface |
| pwiz-sharp/i-agree-to-the-vendor-licenses.bat | Add vendor license opt-in helper |
| pwiz-sharp/build/VendorPinsGenerator/VendorPinsGenerator.csproj | Add build-time tool (net10) |
| pwiz-sharp/build/AgilentPatcher/AgilentPatcher.csproj | Add build-time patcher (net10) |
| pwiz-sharp/.gitignore | Add pwiz-sharp ignores |
| pwiz-sharp/.config/dotnet-tools.json | Add dotCover tool manifest |
| pwiz_tools/Skyline/Util/UtilUI.cs | Use shell-safe opener for URLs/files |
| pwiz_tools/Skyline/Util/Util.cs | Add IList enumerator helper |
| pwiz_tools/Skyline/Util/ScreenCapture.cs | Switch to ControlUtil marshaling |
| pwiz_tools/Skyline/Util/PeptideToMoleculeTextMapper.cs | Update culture code to zh-Hans |
| pwiz_tools/Skyline/Util/PanoramaPublishUtil.cs | Use shell-safe opener for URLs |
| pwiz_tools/Skyline/Util/Install.cs | Fix 64-bit detection on net8+ |
| pwiz_tools/Skyline/Util/CreateHandleDebugBase.cs | Update menu-handle logic for net8+ |
| pwiz_tools/Skyline/ToolsUI/ToolOptionsUI.cs | gRPC channel type update + marshaling |
| pwiz_tools/Skyline/ToolsUI/NativeFolderBrowserDialog.cs | Document classic folder picker constraint |
| pwiz_tools/Skyline/ToolsUI/ConfigureToolsDlg.cs | Force classic folder picker |
| pwiz_tools/Skyline/TestUtil/LongWaitDialogCanceler.cs | Switch to ControlUtil marshaling |
| pwiz_tools/Skyline/TestUtil/KoinaTestUtil.cs | gRPC channel type update |
| pwiz_tools/Skyline/TestUtil/ExtensionTestContext.cs | Use TestRunDirectory for MSTest v3 |
| pwiz_tools/Skyline/TestTutorial/AuditLogTutorialTest.cs | Stabilize tests via WaitForOpenForm |
| pwiz_tools/Skyline/TestRunner/TestRunner32.bat | Batch formatting/line-ending related adjustment |
| pwiz_tools/Skyline/TestPerf/FeatureDetectionTest.cs | Adjust tolerances for net8+/float behavior |
| pwiz_tools/Skyline/TestPerf/AlphapeptdeepBuildLibraryTest.cs | Make sorting deterministic across machines |
| pwiz_tools/Skyline/TestFunctional/SynchSiblingsSmallMoleculesTest.cs | Fix adduct equality expectation under MSTest v3 |
| pwiz_tools/Skyline/TestFunctional/StartPageTest.cs | Use TestRunDirectory for MSTest v3 |
| pwiz_tools/Skyline/TestFunctional/SkylineMcpTest.cs | Bump expected tool ZIP version |
| pwiz_tools/Skyline/TestFunctional/PeakScoringModelTest.cs | Adjust Assert overload behavior under MSTest v3 |
| pwiz_tools/Skyline/TestFunctional/PasteTransitionListTest.cs | Adjust Assert overload behavior under MSTest v3 |
| pwiz_tools/Skyline/TestFunctional/MiscFormsTest.cs | Skip newer MessageBoxButtons values (net472 parity) |
| pwiz_tools/Skyline/TestFunctional/ListClusteringTest.cs | Use TestRunDirectory for MSTest v3 |
| pwiz_tools/Skyline/TestFunctional/LibraryBuildTest.cs | Validate folder-picker automation result |
| pwiz_tools/Skyline/TestFunctional/ImmediateWindowWarningsTest.cs | Fix float formatting expectations across frameworks/locales |
| pwiz_tools/Skyline/TestFunctional/HelpDocumentationContentTest.cs | Update culture code to zh-Hans |
| pwiz_tools/Skyline/TestFunctional/FilesTreeFormTest.cs | Use Enumerable.Reverse to avoid Span.Reverse confusion |
| pwiz_tools/Skyline/TestFunctional/ExportSpectralLibraryTest.cs | Fix SQLite bool parsing on net8+ |
| pwiz_tools/Skyline/TestFunctional/ExportDiaListTest.cs | Use TestRunDirectory for MSTest v3 |
| pwiz_tools/Skyline/TestFunctional/ExportChromatogramTest.cs | Add relative tolerances for net8+/float drift |
| pwiz_tools/Skyline/TestFunctional/ExplicitVariableTest.cs | Use TestRunDirectory for MSTest v3 |
| pwiz_tools/Skyline/TestFunctional/EncyclopeDiaPeptidesWithNoSignalTest.cs | Adjust Assert overload behavior under MSTest v3 |
| pwiz_tools/Skyline/TestFunctional/ConfigureToolsDlgTest.cs | Update ToolMenuItem property name |
| pwiz_tools/Skyline/TestFunctional/AuditLogSavingTest.cs | Use SHA1.Create() for net8+ |
| pwiz_tools/Skyline/TestFunctional/AssayLibraryImportTest.cs | Fix double formatting expectation under net8+ |
| pwiz_tools/Skyline/TestFunctional/AccessServerTest.cs | Adjust Assert overload behavior under MSTest v3 |
| pwiz_tools/Skyline/TestData/ResourcesTest.cs | Include CommonBaseUI/CommonUtil assemblies in resource scan |
| pwiz_tools/Skyline/Test/UniModStaticList.xml | Normalize float serialization |
| pwiz_tools/Skyline/Test/TransitionSettings07Test.cs | Adjust Assert overload behavior under MSTest v3 |
| pwiz_tools/Skyline/Test/SpectrumClassFilterTest.cs | Use Enumerable.Reverse to avoid Span.Reverse confusion |
| pwiz_tools/Skyline/Test/MSstats/Normalization/MsStatsNormalizationTest.cs | Switch to tolerant compare for net8+ formatting |
| pwiz_tools/Skyline/Test/LibraryRankedSpectrumInfoTest.cs | Use Enumerable.Reverse to avoid Span.Reverse confusion |
| pwiz_tools/Skyline/Test/IonMobilityUnitTest.cs | Adjust Assert overload behavior under MSTest v3 |
| pwiz_tools/Skyline/Test/IdentityTest.cs | Adjust Assert overload behavior under MSTest v3 |
| pwiz_tools/Skyline/Test/FindNodeTest.cs | Use TestRunDirectory for MSTest v3 |
| pwiz_tools/Skyline/Test/EspTest.cs | Allow numeric precision diffs in compares |
| pwiz_tools/Skyline/Test/DiskSpaceTest.cs | Use TestRunDirectory for MSTest v3 |
| pwiz_tools/Skyline/Test/CommandLineThermoMethodTest.cs | Update expected command path tail on net8+ |
| pwiz_tools/Skyline/Test/ChromatogramExporterTest.cs | Add relative tolerances for net8+/float drift |
| pwiz_tools/Skyline/Test/AlphapeptdeepLibraryBuilderTest.cs | Add tolerance for formatting differences |
| pwiz_tools/Skyline/SkylineTool/RemoteBase.cs | Silence BinaryFormatter obsoletion warnings |
| pwiz_tools/Skyline/SkylineTool/IJsonToolService.cs | Update language code documentation to zh-Hans |
| pwiz_tools/Skyline/SkylineTester/TabForms.cs | Load Skyline module dll on net8+ apphost change |
| pwiz_tools/Skyline/SkylineTester/AboutWindow.cs | UseShellExecute for URL on net8+ |
| pwiz_tools/Skyline/SkylineNightlyShim/Program.cs | Update renamed ZIP dependency |
| pwiz_tools/Skyline/SkylineNightly/SkylineNightly.skytr | Add new SkylineNightly options |
| pwiz_tools/Skyline/SkylineNightly/SkylineNightly.cs | Force classic folder picker behavior |
| pwiz_tools/Skyline/SkylineNightly/LogFileMonitor.cs | Update ClrMD attach API usage |
| pwiz_tools/Skyline/SkylineNet8Stubs.cs | Add ClickOnce stubs for net8+/net10 |
| pwiz_tools/Skyline/SkylineGraphs.cs | Add alias for ReplicateValue |
| pwiz_tools/Skyline/Skyline.resx | Remove problematic dockPanel size/anchor resources |
| pwiz_tools/Skyline/Skyline.ja.resx | Remove problematic dockPanel size/anchor resources |
| pwiz_tools/Skyline/Skyline.Designer.cs | Dock dockPanel fill to stabilize layout |
| pwiz_tools/Skyline/SettingsUI/ViewLibraryPepInfoList.cs | Disambiguate Range type |
| pwiz_tools/Skyline/SettingsUI/BuildLibraryDlg.cs | Force classic folder picker |
| pwiz_tools/Skyline/Model/Tools/ToolsResources.resx | Add new localized string |
| pwiz_tools/Skyline/Model/Tools/ToolsResources.designer.cs | Designer update for new resource |
| pwiz_tools/Skyline/Model/SearchTool.cs | Add MSAmanda enum value |
| pwiz_tools/Skyline/Model/Results/Scoring/TargetDecoyGenerator.cs | Normalize negative zero formatting |
| pwiz_tools/Skyline/Model/Results/NumericsLsSolver.cs | Remove explicit MathNet provider override |
| pwiz_tools/Skyline/Model/Results/ChromHeaderInfo.cs | Whitespace-only adjustment |
| pwiz_tools/Skyline/Model/Lists/ListItemTypes.cs | Update dynamic assembly API for net8+ |
| pwiz_tools/Skyline/Model/Lib/EncyclopeDiaHelpers.cs | Resolve bundled msconvert exe path |
| pwiz_tools/Skyline/Model/Lib/ChromLib/Data/mapping.xml | Fix NHibernate mapped id properties |
| pwiz_tools/Skyline/Model/Lib/BiblioSpecLite.cs | Close redundant connection earlier for net8+ file sharing |
| pwiz_tools/Skyline/Model/Koina/Config/KoinaConfig.cs | Switch to Grpc.Net.Client channel |
| pwiz_tools/Skyline/Model/Koina/Communication/KoinaPredictionClient.cs | Accept ChannelBase for new channel type |
| pwiz_tools/Skyline/Model/ImportPeakBoundaries.cs | Fix loop variable shadowing/logic |
| pwiz_tools/Skyline/Model/DdaSearch/MsconvertDdaConverter.cs | Resolve bundled msconvert exe path |
| pwiz_tools/Skyline/Model/DdaSearch/DdaSearchResources.resx | Add MSAmanda/Percolator messages |
| pwiz_tools/Skyline/Model/AuditLog/BlockHash.cs | Use SHA1.Create() for net8+ |
| pwiz_tools/Skyline/Model/AbstractMassListExporter.cs | Persist-round m/z for byte-identical exports |
| pwiz_tools/Skyline/Menus/ViewMenu.cs | Add Koina namespace import |
| pwiz_tools/Skyline/Menus/PeakAreasContextMenu.cs | Add alias for ReplicateValue |
| pwiz_tools/Skyline/Menus/EditMenu.cs | Add alias for ReplicateValue |
| pwiz_tools/Skyline/Menus/ContextMenuControl.cs | Add alias for ReplicateValue |
| pwiz_tools/Skyline/FileUI/PublishDocumentDlgArdia.cs | Switch to ControlUtil marshaling |
| pwiz_tools/Skyline/FileUI/PeptideSearch/SearchSettingsControl.cs | Hook required files for MSAmanda |
| pwiz_tools/Skyline/FileUI/PeptideSearch/ImportResultsControl.cs | Force classic folder picker |
| pwiz_tools/Skyline/FileUI/PeptideSearch/EncyclopeDiaSearchDlg.cs | Use MD5.Create() for net8+ |
| pwiz_tools/Skyline/FileUI/ImportResultsDlg.cs | Force classic folder picker |
| pwiz_tools/Skyline/Executables/Tools/ToolServiceTestHarness/ToolServiceTestHarness.csproj | Update TFM to net10.0-windows |
| pwiz_tools/Skyline/Executables/Tools/SortProteinsTool/SortProteins/SortProteins.csproj | Update TFM to net10.0 |
| pwiz_tools/Skyline/Executables/Tools/SkylineMcp/SkylineAiConnector/tool-inf/info.properties | Update required runtime text to .NET 10 |
| pwiz_tools/Skyline/Executables/Tools/SkylineMcp/SkylineAiConnector/MainForm.cs | Update runtime check path and messaging |
| pwiz_tools/Skyline/Executables/Tools/AdvancedEditingCommands/AdvancedEditingCommands.csproj | Update TFM to net10.0-windows |
| pwiz_tools/Skyline/Executables/SkylineBatch/SkylineBatchTest/SkylineSettingsTest.cs | Test infrastructure tweak for async deadlock |
| pwiz_tools/Skyline/Executables/SkylineBatch/SkylineBatchTest/RemoteFileSourceFunctionalTest.cs | Add Panorama HTTP mock behavior |
| pwiz_tools/Skyline/Executables/SkylineBatch/SkylineBatchTest/BcfgFileTest.cs | Rewrite SkylineCmd path for net8+ layout |
| pwiz_tools/Skyline/Executables/SkylineBatch/SkylineBatchTest/AbstractSkylineBatchFunctionalTest.cs | Ensure HttpClient test behavior cleanup |
| pwiz_tools/Skyline/Executables/SkylineBatch/SkylineBatch/RDirectorySelector.cs | Force classic folder picker |
| pwiz_tools/Skyline/Executables/SharedBatch/SharedBatch/SkylineTypeControl.cs | Force classic folder picker |
| pwiz_tools/Skyline/Executables/SharedBatch/SharedBatch/ProcessRunner.cs | Close stdin to avoid net8+ hang |
| pwiz_tools/Skyline/Executables/SharedBatch/SharedBatch/FindSkylineForm.cs | Force classic folder picker |
| pwiz_tools/Skyline/Executables/SharedBatch/SharedBatch/FileUtil.cs | Force classic folder picker |
| pwiz_tools/Skyline/Executables/Installer/FileList64-template.txt | Swap PortableUtil -> CommonBaseUI artifacts |
| pwiz_tools/Skyline/Executables/DevTools/Utf16to8/Utf16to8.csproj | Update TFM to net10.0 |
| pwiz_tools/Skyline/Executables/DevTools/ResourcesOrganizer/Test/Test.csproj | Update TFM to net10.0 |
| pwiz_tools/Skyline/Executables/DevTools/ResourcesOrganizer/ResourcesOrganizer/ResourcesOrganizer.csproj | Update TFM to net10.0 |
| pwiz_tools/Skyline/Executables/DevTools/NormalizeResxWhitespace/NormalizeResxWhitespace.csproj | Update TFM to net10.0 |
| pwiz_tools/Skyline/Executables/DevTools/ImageExtractor/ImageExtractor.csproj | Update TFM to net10.0-windows |
| pwiz_tools/Skyline/Executables/DevTools/ImageConverter/ImageConverter/ImageConverter.csproj | Update TFM to net10.0-windows |
| pwiz_tools/Skyline/Executables/DevTools/ImageComparer/ImageComparerWindow.cs | Force classic folder picker behavior |
| pwiz_tools/Skyline/Executables/DevTools/ImageComparer.Mcp/ImageComparer.Mcp.csproj | Update TFM to net10.0-windows |
| pwiz_tools/Skyline/Executables/DevTools/ImageComparer.Core/ImageComparer.Core.csproj | Update TFMs; Windows-only System.Drawing for net10 |
| pwiz_tools/Skyline/Executables/DevTools/AssortResources/CsProjFile.cs | Fix root-level resource path combine |
| pwiz_tools/Skyline/Executables/BuildMethod/BuildWatersMethod/Program.cs | Remove obsolete remoting using |
| pwiz_tools/Skyline/Executables/AutoQC/AutoQC/AutoQCConfigForm.cs | Force classic folder picker |
| pwiz_tools/Skyline/EditUI/ComparePeakPickingDlg.cs | Null-guard scaler on early resize |
| pwiz_tools/Skyline/Controls/Startup/ActionTutorial.cs | Use shell-safe opener for PDF |
| pwiz_tools/Skyline/Controls/Graphs/RTScheduleGraphPane.cs | Switch to pwiz-sharp Bruker namespace |
| pwiz_tools/Skyline/Controls/Graphs/LabelLayoutRunner.cs | Switch to ControlUtil marshaling |
| pwiz_tools/Skyline/Controls/Graphs/GraphSpectrum.cs | Switch to ControlUtil marshaling |
| pwiz_tools/Skyline/Controls/Graphs/GraphDataCalculator.cs | Switch to ControlUtil marshaling |
| pwiz_tools/Skyline/Controls/Graphs/ExportMethodScheduleGraph.cs | Fix unchecked enum cast + namespace |
| pwiz_tools/Skyline/Controls/Graphs/AreaAbundanceComparisonGraphPane.cs | Update caching namespace |
| pwiz_tools/Skyline/Controls/FilesTree/BackgroundActionService.cs | Switch to ControlUtil marshaling |
| pwiz_tools/Skyline/CommandArgs.cs | Formatting + culture arg text update |
| pwiz_tools/Skyline/Alerts/ShareResultsFilesDlg.cs | Force classic folder picker |
| pwiz_tools/Skyline/Alerts/PathChooserDlg.cs | Force classic folder picker |
| pwiz_tools/Skyline/Alerts/ArdiaLoginDlg.cs | Loosen CoreWebView2 exception wrapper type |
| pwiz_tools/Skyline/.config/dotnet-tools.json | Add dotCover tool manifest |
| pwiz_tools/Shared/zedgraph/ZedGraph/ZedGraphException.cs | Suppress obsolete serialization warning |
| pwiz_tools/Shared/zedgraph/ZedGraph/ZedGraphControl.Events.cs | UseShellExecute for URL on net8+ |
| pwiz_tools/Shared/zedgraph/ZedGraph.csproj.user | Mark ZedGraph controls as UserControl |
| pwiz_tools/Shared/ProteowizardWrapper.PwizSharp/MsDataFileInfo.cs | Add pwiz-sharp wrapper helper |
| pwiz_tools/Shared/ProteowizardWrapper.PwizSharp.Smoke/ProteowizardWrapper.PwizSharp.Smoke.csproj | Add smoke project for wrapper |
| pwiz_tools/Shared/ProteomeDb/Util/DatabaseResource.cs | Recreate stale closed session factory |
| pwiz_tools/Shared/PanoramaClient/PanoramaFilePicker.cs | Use shell-safe opener for URLs |
| pwiz_tools/Shared/PanoramaClient/PanoramaDirectoryPicker.cs | Use shell-safe opener for URLs |
| pwiz_tools/Shared/Lib/Microsoft.VC90.CRT/x86/Microsoft.VC90.OpenMP.manifest | Add manifest elements |
| pwiz_tools/Shared/Lib/Microsoft.VC90.CRT/x64/Microsoft.VC90.OpenMP.manifest | Add manifest elements |
| pwiz_tools/Shared/CommonUtil/SystemUtil/ProcessEx.cs | Add OpenInShell helper |
| pwiz_tools/Shared/CommonUtil/SystemUtil/CommonException.cs | Suppress obsolete serialization warning |
| pwiz_tools/Shared/CommonUtil/SystemUtil/CommandStatusWriter.cs | Avoid StringBuilder WriteLine overload pitfall |
| pwiz_tools/Shared/CommonUtil/SystemUtil/Caching/Producer.cs | Remove WinForms-dependent helper |
| pwiz_tools/Shared/CommonUtil/SystemUtil/Assume.cs | Use shell-safe opener for .sln |
| pwiz_tools/Shared/CommonUtil/CommandLine/NameValuePair.cs | Restore invalid-path rejection under net8+ |
| pwiz_tools/Shared/CommonUtil/CommandLine/IUsageBlock.cs | Add usage block interface |
| pwiz_tools/Shared/CommonUtil/Collections/LinqExtensions.cs | Remove WinForms SortOrder overload helper |
| pwiz_tools/Shared/CommonMsData/RemoteApi/WatersConnect/WatersConnectAccount.cs | Surface raw identity server errors |
| pwiz_tools/Shared/CommonMsData/RemoteApi/RemoteServerException.cs | Suppress obsolete serialization warning |
| pwiz_tools/Shared/CommonMsData/RemoteApi/RemoteAccountType.cs | Formatting and using order |
| pwiz_tools/Shared/CommonFileDialogs/BaseFileDialogNE.cs | Improve single-account navigation behavior |
| pwiz_tools/Shared/CommonBaseUI/SystemUtil/FormUtil.cs | Add classic FolderBrowserDialog factory |
| pwiz_tools/Shared/CommonBaseUI/GUI/CommonAlertDlg.cs | Use Enumerable.Reverse to avoid Span.Reverse confusion |
| pwiz_tools/Shared/Common/Directory.Build.targets | Suppress CA1416 only for Windows-targeted projects |
| pwiz_tools/Shared/Common/Directory.Build.props | Add NoWarn/MSBuildWarningsAsMessages tuning for net10 |
| pwiz_tools/Shared/Common/DataBinding/DsvWriter.cs | Round-trip formatting compatibility |
| pwiz_tools/Shared/Common/DataBinding/DataSchema.cs | Restore Uri text filtering behavior |
| pwiz_tools/Shared/Common/DataBinding/Controls/Editor/DocumentationViewer.cs | Switch to ControlUtil marshaling |
| pwiz_tools/Shared/Common/Database/NHibernate/StatelessSessionWithLock.cs | Suppress obsolete interface member warning |
| pwiz_tools/Shared/Common/DataAnalysis/alglib/alglib_info.cs | Remove third-party assembly identity attributes |
| pwiz_tools/Osprey/tcbuild.bat | Build net10.0 + dotCover from tool manifest |
| pwiz_tools/Osprey/Regression/BlibGolden.ps1 | Update net10.0 paths/messages |
| pwiz_tools/Osprey/regression.ps1 | Update net10.0 paths/messages |
| pwiz_tools/Osprey/Osprey/Osprey.csproj | Switch project reference to CommonUtil |
| pwiz_tools/Osprey/Osprey.Scoring/Osprey.Scoring.csproj | Comment update for transitive refs |
| pwiz_tools/Osprey/Osprey.IO/MzmlReader.cs | Consolidate XML double parsing path |
| pwiz_tools/Osprey/Osprey.Core/Osprey.Core.csproj | Remove net472-only System.Memory reference |
| pwiz_tools/Osprey/Directory.Build.targets | Update wording for net10.0 |
| pwiz_tools/Osprey/Directory.Build.props | Switch Osprey to net10.0 only |
| pwiz_tools/Osprey/.config/dotnet-tools.json | Add dotCover tool manifest |
| pwiz_tools/BiblioSpec/CleanBiblioSpec.bat | Batch formatting/line-ending related adjustment |
| pwiz_aux/msrc/utility/vendor_api/Bruker/Baf2Sql.cpp | Add instrument family mapping |
| libraries/SQLite/update3rdPartyDLLs.bat | Batch formatting/line-ending related adjustment |
| global.json | Pin repo SDK to .NET 10 |
| .gitignore | Ignore additional net8+/net10 test artifacts |
| .gitattributes | Enforce CRLF for .bat; preserve fixture bytes |
| .editorconfig | Demote WinForms WFO1000 to warning |
Review details
- Files reviewed: 42/1951 changed files
- Comments generated: 5
- Review effort level: Lite
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
Comment on lines
166
to
175
| var result = MessageBox.Show(this, | ||
| "The AI Connector requires the .NET 8.0 Desktop Runtime, which was not found on this computer.\n\n" + | ||
| "The AI Connector requires the .NET 10.0 Desktop Runtime, which was not found on this computer.\n\n" + | ||
| "Would you like to open the download page?", | ||
| ".NET 8.0 Required", | ||
| ".NET 10.0 Required", | ||
| MessageBoxButtons.YesNo, | ||
| MessageBoxIcon.Warning); | ||
| if (result == DialogResult.Yes) | ||
| Process.Start(McpServerDeployer.DotNetDownloadUrl); | ||
| labelStatus.Text = ".NET 8.0 Desktop Runtime is required."; | ||
| labelStatus.Text = ".NET 10.0 Desktop Runtime is required."; | ||
| return; |
Member
Author
There was a problem hiding this comment.
Not addressed in this round, leaving open. The policy is right, but this file already has 39 hardcoded user-facing strings, so moving these three into .resx in isolation would leave the rest inconsistent. Localizing SkylineAiConnector properly is worth doing as its own change rather than as part of a review round.
Comment on lines
+51
to
+52
| if ((int)buttonSet > (int)MessageBoxButtons.RetryCancel) | ||
| continue; |
Member
Author
There was a problem hiding this comment.
Pushing back on this one, leaving open for a human call. ai/CRITICAL-RULES.md says: "If statements must not be single-line. Keep condition and body on separate lines if braces omitted." The condition and continue; are already on separate lines, so this follows the rule as written, and the surrounding code uses the same form. Happy to add braces if the guideline is meant to be stricter than it reads.
* SkylineAiConnector: opening the .NET download page used Process.Start's string overload, which on .NET Core and later starts an executable rather than shell-executing a URL, so answering Yes threw Win32Exception. Passed a ProcessStartInfo with UseShellExecute instead. The prompt appears exactly when the runtime is missing, so this is the path that most needs to work. ProcessEx.OpenInShell would have been the natural helper, but this tool references only SkylineTool and ships as its own zip. * pwiz-sharp README said .NET 8 / net8.0 in two places; the projects target net10.0. Not addressed, left open for review: the hardcoded strings in the same MainForm.cs prompt (the file has 39 of them, so localizing three in isolation would be inconsistent), and the single-line if in MiscFormsTest.cs, which keeps condition and body on separate lines as CRITICAL-RULES allows. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0182oRXbeKQGEpF1kTkdQAsr
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Adds
pwiz-sharp/, a .NET 10.0 C# port of the ProteoWizard core, and wires it into TeamCity via the newProteoWizard_CoreWindowsNetbuild config.The port currently covers:
Pwiz.Util,Pwiz.Data.Common,Pwiz.Data.MsData(mzML / mzXML / MGF read+write, indexed envelopes, SHA-1 footers, ParamContainer / CV term machinery, lossy round-trip diff)src/Vendor/): Thermo (RawFileReader), Bruker (BAF / TSF / TDF + PASEF, plus YEP / FID through CompassXtract — no Bruker format is now unported), Waters (MassLynx + IMS / MSe / SONAR / lockmass), Agilent (Cecil-patched on .NET 8 to removeDelegate.BeginInvoke; MS scans + IM frames + UV/DAD + SRM/SIM transitions + pump/sampler chromatograms), Sciex (.wiffvia legacy AnalystDataProvider,.wiff2viaSCIEX.Apis.Data.v1in a side-by-sideAssemblyLoadContext), Shimadzu, UNIFI, UIMF, Mobilion, Bruker PRM schedulingmsconvert-sharp: command-line entry point exercising the full stackcpp parity
The port is measured against C++
msconvertover a 492-file corpus (every vendor file under 100 MB in the shared test tree), converting both with identical flags and comparing the mzML semantically rather than byte-wise."Current" is the 12 Aug sweep plus 25 files individually re-verified against C++ since it ran (Sciex 7, Agilent 8, Waters 6, Thermo 4). Per vendor: Sciex 180/226, Thermo 93/109, Agilent 33/50, Waters 32/56, Bruker 17/20, Mobilion 3/7, PerkinElmer 1/1, Shimadzu 0/23.
cpp-failed(35) andboth-failed(33) are files C++ itself cannot convert; they are excluded from the comparison rather than counted against the port.Differences that are not port defects
Roughly half the remaining 62 are blocked on something other than our code, and are documented at the relevant call sites:
1261.77vs1261.77332). Verified by running the C# binary against cpp's 4.9 DLL, which reproduces cpp exactly. A 5.0 upgrade for the cpp tree is staged on a branch; note 5.0 drops thesetRawPathexport and theMassLynxRawProcessorheader, so it is a small API migration rather than a DLL swap.SampleInfo.AnalysisDateatMinValueunder .NET 8, sostartTimeStampfalls back toFilePropTag.GeneratedDateTime, 15 s away. Not BinaryFormatter —EnableUnsafeBinaryFormatterSerializationis already set and the field is still empty.MS:1000803 analyzer scan offsetfor PDA scans from uninitialized memory: two consecutive runs of the same binary emit 689 of them, then 122. Deliberately not reproduced.598.125→598.13vs598.12). Both sides call the same SDK method.GetChromatogramto subsequent spectrum reads (~5e-10 in the m/z axis). No single reader state satisfies both the files that need the chromatogram-sourced TIC and the files that need an unperturbed m/z axis; the safer half of the trade is taken and documented.Recent parity work
Vendor-reader fixes landed since the last sweep, each verified by converting the affected corpus files with both binaries:
GetSignalInfo); TIC intensities rounded throughfloatas cpp caches them; spectrum TIC / base-peak taken from the cached chromatograms rather than the scan record (they disagree by up to 12% on centroided QTOF runs)IC{n}with its param-group and software refs; instrument model resolved from the method text (cpp's first source, which we lacked); TIC ms-level array written as the rawMSOrder(negative for neutral-loss/parent scans); CAD intensities scaled by 1e-6ReaderConfig.VerifyNonEmptySpectraAtIndex, on for conversion, off for interactive callers)--srmAsSpectramade functional (it was hardcoded off, so the flag silently produced 15 spectra where cpp produced 5,112; the other four vendors were checked and were already correct)ToUniversalTimecall); emptyspectrumListomitted$(OutputPath)wiff2was string-concatenated, sodotnet build -o <dir>staged the wiff2 plugin into a sibling directory and every.wiff2silently failed to openPort status
Auto-generated from
pwiz/**/Jamfile.jamandpwiz_tools/{commandline,MSConvertGUI,SeeMS}/Jamfile.jam— each box is a pwiz module directory; arrows aggregate at the directory level after transitive reduction. Generator script:pwiz-sharp/docs/jamdep.py.Legend — green = ported and parity-checked, yellow = ported with known gaps, red = not started.
Notes on partials
data/msdataReaderConfig.RunIndexround-trips but every reader still loads run 0 (SeeMS already parsespath:N); IndexedMzMLBuffer optimisationsdata/identdatadata/tradataanalysis--filtercoverage is not exhaustive against cpp's setanalysis/.../LockmassRefinerSpectrumList_LockmassRefiner+ChromatogramListLockmassRefinerunderPwiz.Analysisvendor_readers/Brukervendor_readers/Sciexno-spectraWIFF still fails to open (STG_E_FILENOTFOUND)vendor_readers/Agilentvendor_readers/Waters--srmAsSpectravendor_readers/Shimadzu--srmAsSpectrastartTimeStampblocked on the SDK (see above)vendor_readers/UNIFI,UIMF,MobilionAbstractWatersHttpReader; not corpus-swept (network / niche formats)utility/proteomeutility/misctools/msconvertmsconvert--filtercompleteness gaps mirror the analysis-module gapstools(other)TeamCity wiring
pwiz-sharp/build.bat— TeamCity entry point. Runsdotnet restore+dotnet build Pwiz.sln+dotnet test Pwiz.sln --no-buildwith##teamcity[progressMessage]markers,--logger:trx, and--logger:teamcity(TeamCity.VSTest.TestAdapter — emits per-test##teamcity[testStarted/Finished/Failed]so individual tests show up in the TC UI). Defaults to Release; passDebugas the first arg to switch.pwiz-sharp/global.jsonpins the SDK to .NET 8 (latestFeature) so dotCover 2023.3.3 (which doesn't support .NET 9/10) keeps working.pwiz-sharp/build/AgilentPatcher/— Mono.Cecil tool that rewrites the Agilent SDK to removeDelegate.BeginInvoke/EndInvokecalls (removed from .NET 5+) and to fix a partial-read bug that made large spectra unreadable on .NET Core. Runs afterExtractAgilentAssemblies; patched DLLs are gitignored.pwiz-sharp/i-agree-to-the-vendor-licenses.bat— writes a per-userDirectory.Build.user.propssettingIAgreeToVendorLicenses=trueso vendor SDKs work in Visual Studio without per-build property injection.scripts/misc/vcs_trigger_and_paths_config.py— added aCoreWindowsNettarget that maps toProteoWizard_CoreWindowsNet, plus apwiz-sharp/.*matchPath placed before the generic.bat/scripts/librariespatterns. Net effect: changes underpwiz-sharp/trigger only the .NET config; cpp changes outside that tree don't trigger the .NET config.Test plan
cd pwiz-sharp && build.bat Debugsucceeds locallyProteoWizard_CoreWindowsNetbuild config exists in TeamCity and runspwiz-sharp/build.bat(ortcbuild.bat)ProteoWizard_CoreWindowsNetpassesProteoWizard_CoreWindowsNetpwiz-sharp/-only change does NOT trigger the cpp Core/Skyline/Bumbershoot chain🤖 Generated with Claude Code